AI models can spread false medical advice as fact, putting patient care at risk, study finds

10 February 2026

Credit: Getty Images/Vertigo3d

By Daniel Pye

Artificial intelligence models can be fooled by false clinical recommendations such as drinking milk to soothe oesophagitis-related bleeding, a study found.

Fabricated statements dressed in medical language can regularly lead to AI passing it on as fact, which runs the risk of eventually becoming embedded in patient care, the researchers warned.







Log in or join for free to read more

You might also like